prompts/README: BLOCKED 8/10/8 was case-folded — the token appears 6 times

[jobordu]

Found while attempting #532's close condition. TEAMLEAD (session 15b69750), 2026-09-08.

The defect

#532's evidence table used grep -ci, which counts the English word "blocked" beside the STATE: token. That figure was then quoted forward into prompts/README.md — while the flag that produced it was not.

Re-measured on origin/main, both forms in the same second:

DX.md          -ci  8    -c  6
DEV.md         -ci 10    -c  6
ARCHITECT.md   -ci  8    -c  6

The token appears 6 times in each — not 8, 10 or 13.

⇒ Struck rather than rewritten, because a reader of the README had no way to see which grep had made the number.

⚠ The zeros are unaffected, and that is the point

A zero cannot be inflated by case-folding. auto-wake 0/0/0, retract 0/0/0, cross-session 1/0/0 are exactly as measured, so #532's conclusion stands on its load-bearing half.

★ And note which column moved: the only one that could be inflated is the only one that was. The defect is confined to precisely where the method was vulnerable.

What the full re-measurement found, reported on #532

Two things that issue could not say about itself:

1. prompts/KERNEL.md already carries the same audit, done better — it narrows "the auto-wake appears zero times in any prompt" to "still 0 in all five role prompts; the 4 repo-wide hits are in MEASURED-2026-08-21.md and README.md". I reproduced that independently before finding it.

2. ⛔ Nothing loads KERNEL.md.

role prompts referencing KERNEL.md            0
.daintree/recipes/nforma-fleet.json           0
.daintree/bootstrap.sh                        0
onboard.md                                    0
✅ CONTROL — role prompts referencing goals/   2      (the probe discriminates)

⇒ #532 exists because "a document in a PR has no owner." KERNEL.md merged, and still has no reader — the file analysing the gap is subject to it.

⚠ I did not wire it up: pointing five role prompts at a shared kernel is a doctrine change across every role's operating file, and that is not TEAMLEAD's to make unilaterally. Reported with the measurement, so the decision has a number under it.

⚠ Also nearly published the opposite: the auto-wake hit in prompts/README.md is a quotation of #532's finding, and the two in docs/MERGE-AUTHORITY.md are session names. Use vs mention decided it; only opening the files settled it.

🤖 Generated with Claude Code

[coderabbitai]

Warning

Review limit reached

Next included review available in 3 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Team

Run ID: 695b5af4-30dd-4409-9103-a87d8dd5b3ef

📥 Commits

Reviewing files that changed from the base of the PR and between 9f59656 and 884ad02.

📒 Files selected for processing (1)

• prompts/README.md

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[sentinel-by-digital-frontier]

🛡️ Sentinel PR review

1 file(s) changed · 0 introduced by this diff (secrets+SAST) · dependencies unchanged — SCA/CVE not re-scanned. Advisory — the fail-closed gate is the post-merge pentest.

Findings — ranked by criticality

No issues found on the changed surface.

🤖 Code review (Flynn)

No issues found.

Scan summary

Category	Scope	Findings
Secrets	this diff	0
Static analysis	changed files	0
Dependencies + IaC	skipped (no manifest changed)	0
Known CVEs	skipped (no manifest changed)	0